Clean up unused Rust dependencies, eliminate duplicate repository

WORKSPACE

@@ -50,15 +50,6 @@ http_archive(
     urls = ["https://github.com/google/boringssl/tarball/f7cf966f3ddc6923104f6a354bf0ba5c618f3320"],
 )
 
-http_archive(
-    name = "com_cloudflare_lol_html",
-    url = "https://github.com/cloudflare/lol-html/tarball/a0053299f6809c2fa4e3af35a4f64bd8069952ba",
-    strip_prefix = "cloudflare-lol-html-a005329",
-    type = "tgz",
-    sha256 = "eba5f6ce291bc0f8e1ba588573c5e88a6a1ba4264b7961b1a674fdbe334b50c2",
-    build_file = "//:build/BUILD.lol-html",
-)
-
 http_archive(
     name = "sqlite3",
     url = "https://sqlite.org/2022/sqlite-amalgamation-3400100.zip",
@@ -128,6 +119,9 @@ http_archive(
 # Rust bootstrap
 #
 # workerd uses some Rust libraries, especially lolhtml for implementing HtmlRewriter.
+# Note that lol_html itself is not included here to avoid dependency duplication and simplify
+# the build process. To update the dependency, update the reference commit in
+# rust-deps/BUILD.bazel and run `bazel run //rust-deps:crates_vendor -- --repin`
 
 http_file(
     name = "cargo_bazel_linux_x64",
@@ -200,10 +194,6 @@ load("//rust-deps/crates:crates.bzl", "crate_repositories")
 
 crate_repositories()
 
-load("//rust-deps/cxxbridge_crates:crates.bzl", cxxbridge_repositories = "crate_repositories")
-
-cxxbridge_repositories()
-
 load("@rules_rust//tools/rust_analyzer:deps.bzl", "rust_analyzer_dependencies")
 
 rust_analyzer_dependencies()
@@ -351,3 +341,15 @@ new_local_repository(
         visibility = ["//visibility:public"])""",
     path = "empty",
 )
+
+# rust-based lolhtml dependency, including the API header. See rust-deps for details.
+new_local_repository(
+    name = "com_cloudflare_lol_html",
+    build_file_content = """cc_library(
+        name = "lolhtml",
+        hdrs = ["@workerd//rust-deps:lol_html_api"],
+        deps = ["@workerd//rust-deps"],
+        strip_include_prefix = "/",
+        visibility = ["//visibility:public"],)""",
+    path = "empty",
+)

rust-deps/BUILD.bazel

@@ -1,4 +1,3 @@
-load("//:build/rust_cxx_bridge.bzl", "rust_cxx_bridge", "rust_cxx_include")
 load("@rules_rust//crate_universe:defs.bzl", "crate", "crates_vendor")
 load("@rules_rust//rust:defs.bzl", "rust_static_library")
 load("@bazel_skylib//lib:selects.bzl", "selects")
@@ -46,21 +45,17 @@ CARGO_BAZEL = select({
     ":win_x64": "@cargo_bazel_win_x64//file:downloaded.exe",
 })
 
-# TODO (before prod):
-# - arm64 build
-# - Move lolhtml back to submodule so update-deps works
-
 # Generates a repository containing all the crates we reference from our
 # rust workspace
 # To repin crates: bazel run //rust-deps:crates_vendor -- --repin
 crates_vendor(
     name = "crates_vendor",
     annotations = {
         "lolhtml": [crate.annotation(
-            shallow_since = "1657163695 -0400",
+            shallow_since = "1686149015 +0100",
         )],
         "lol_html": [crate.annotation(
-            shallow_since = "1657163695 -0400",
+            shallow_since = "1686149015 +0100",
         )],
     },
     cargo_bazel = CARGO_BAZEL,
@@ -70,13 +65,6 @@ crates_vendor(
             features = ["default"],
             version = "1",
         ),
-        "cxx": crate.spec(
-            features = ["default"],
-            version = "1",
-        ),
-        "typed-arena": crate.spec(
-            version = "2",
-        ),
         "libc": crate.spec(
             version = "0.2",
         ),
@@ -91,25 +79,9 @@ crates_vendor(
             ],
             version = "0.14",
         ),
-        "object": crate.spec(
-            default_features = False,
-            # Only need to be able to read ELF files.
-            features = [
-                "std",
-                "read_core",
-                "elf",
-                "archive",
-                "compression",
-                "unaligned",
-            ],
-            version = "0.23",
-        ),
-        "backtrace": crate.spec(
-            version = "0.3",
-        ),
         "lolhtml": crate.spec(
             git = "https://github.com/cloudflare/lol-html.git",
-            rev = "a0053299f6809c2fa4e3af35a4f64bd8069952ba",
+            rev = "2681dcf0b3e6907111565199df8c43cc9aab7fe8",
         ),
 
         # Used for RTTI parameter extraction
@@ -144,74 +116,29 @@ crates_vendor(
     ],
 )
 
-# To repin crates: bazel run //rust-deps:cxxbridge_vendor -- --repin
-crates_vendor(
-    name = "cxxbridge_vendor",
-    cargo_bazel = CARGO_BAZEL,
-    mode = "remote",
-    packages = {
-        "cxxbridge-cmd": crate.spec(
-            version = "1.0.72",
-        ),
-    },
-    # host toolchain only
-    supported_platform_triples = [
-        "aarch64-apple-darwin",
-        "x86_64-apple-darwin",
-        "aarch64-unknown-linux-gnu",
-        "x86_64-unknown-linux-gnu",
-    ],
-    vendor_path = "cxxbridge_crates",
-)
-
-rust_cxx_include(
-    name = "cxx/include",
-    include_prefix = "rust",
-    visibility = ["//visibility:public"],
-)
-
-rust_cxx_bridge(
-    name = "addr2line",
-    src = "src/addr2line.rs",
-    include_prefix = "workerd/rust",
-    strip_include_prefix = "src",
-    visibility = ["//visibility:public"],
-)
-
-rust_cxx_bridge(
-    name = "util",
-    src = "src/util.rs",
-    include_prefix = "workerd/rust",
-    strip_include_prefix = "src",
-    visibility = ["//visibility:public"],
-)
-
 rust_static_library(
     name = "rust-deps",
-    srcs = glob(["**/*.rs"]),
+    srcs = ["src/lib.rs"],
     stamp = -1, # default to bazel --stamp flag
     # When stamping is enabled this will be replaced by the corresponding
     # value in ./bazel-out/volatile-status.txt
     rustc_env = {
         "WORKERD_VERSION": "{WORKERD_VERSION}"
     },
     visibility = ["//visibility:public"],
-    deps = select({
+    deps = [
         # On Windows, CXX is broken under Bazel (https://github.com/dtolnay/cxx/pull/125).
         # Luckily, util.rs and addr2line.rs are only used in the internal build which just targets
         # Linux, so we can safely disable compilation of any CXX-using code.
-        "@platforms//os:windows": ["@crates_vendor//:lolhtml"],
-        "//conditions:default": [
-            ":addr2line",
-            ":util",
-            "@crates_vendor//:addr2line",
-            "@crates_vendor//:anyhow",
-            "@crates_vendor//:backtrace",
-            "@crates_vendor//:cxx",
-            "@crates_vendor//:libc",
-            "@crates_vendor//:lolhtml",
-            "@crates_vendor//:object",
-            "@crates_vendor//:typed-arena",
-        ]
-    }),
+        "@crates_vendor//:lolhtml",
+    ],
+)
+
+# define the path to the lolhtml headers, this is clunky but allows us to avoid cloning the
+# repository twice.
+alias(
+    name = "lol_html_api",
+    actual = "@crates_vendor__lol_html-1.0.1//:c-api/include/lol_html.h",
+    tags = ["manual"],
+    visibility = ["//visibility:public"],
 )